I hugely prefer w3m's whitespace between paragraphs. Lynx's formatting might be better-suited to 80x24 terminals, but at larger sizes, w3m (or the links/elinks browsers) are much easier reading.
I love using Lynx on medium.com. It strips away all the non-content and leaves the good stuff. In fact, I can't browse medium pages using any other browser.
Reader mode Firefox is my tool for that (on mobile too - never "upgraded" to Fenix).
I often open a news page or something and it's completely blank at first because so many scripts and assets got blocked by uMatrix. Before I actually unblock any domains, I try toggling reader mode on - a lot of times it's enough to let me read the plain HTML text which I actually wanted.
Oh, that's great to hear! Still not going to upgrade unless some extremely critical vulnerability comes out, though. Tab queue and uMatrix are both killer features for me.
I quit using Lynx (and other text mode browsers) when I started hitting a lot of sites that wouldn't render. This was largely after Flash. I seem to remember JavaScript being a problem. I started using Lynx in the 90's over dialup, telneting into a Unix machine.
Sounds about like my experience too. I used to use it about half of the time up until 2003 or so when I got tired of having to guess how much hassle it was going to be to deal with frames, etc.
I use it regularly, mostly to reading articles on news websites which are getting more and more cluttered with heavy video-based ads and all sorts of garbage.
But I used to use it more often. Some years ago I worked mostly on a terminal window and by using lynx, I could browse the internet while everybody thought I was doing some serious work on the server (folks over there respected people working on black text-based screens)
Reader mode in Firefox was a terrific idea and I use it daily. It's "disable images" on steroids. I can't recommend it enough. I do wish there was a simple "disable images" button for other instances when reader mode wasn't available.
I believe Lynx does have a vi mode you can enable, though I might be thinking of w3m...
But the killer app for me with web browsers is link hints, like in several browser extensions (e.g. vrome, tridactyl) and in browsers like qutebrowser. It takes a little getting used to at first to use efficiently, but I hate not having it when I use Lynx or w3m.
Slightly apropos, and for what it's worth, I've noticed that 99% of my browsing seems to be or start from maybe 20 or so web sites. I've started creating shell aliases to open those sites and there is no reason not to use the best browser for the site for each one. It works well with a philosophy like seems to be the one behind the surf browser, where you open each thing in its own window and use your window manager to deal with it, instead of using tabs.
Different browsers for different stuff is why I made a 'router' of sorts for my browsers. Hoping to product-ize it someday, but so many browser quirks! (There are actually many of these switchers but none which try to transparently intercept navigation in browser.)
As someone who uses Elinks/Links a lot, I'd highly recommend switching out Elinks with Links as the former is a now-deprecated fork of the latter [0][1]
I've been using lynx for almost 3 decades. It was my primary browser for over a decade. The modern web has unfortunately left it behind but it still has utility. I have used it to test accessibility on sites. It is an excellent file manager. Just bookmark ~/ for navigation. It is useful for importing pages with -dump from the command line. It is arguably the best ever gopher client. It is also configurable beyond belief. Hopefully with a resurgence of interest in simplicity and text it will continue to be useful...
links and elinks are actually two different browsers.
elinks started out as an experimental fork of links but these days they’ve diverged a little. For example, links supports a whole plethora of operating systems, including DOS! While elinks is POSIX only.
Bias disclosure: I have been a links user for over 20 years. It is my primary browser for recreational web use. As an internet user beginning the late 80's, I always tried every linemode or text-only browser I heard about. I think I have possibly used them all. I used lynx through a public access UNIX system for some years. It baffles me when people recommend lynx today. It makes me think they do not actually use a text-only browser much (e.g., once in a while), and that they have not tried them all. To answer your question, the two biggest reasons I would always choose links over lynx is 1. the way links renders tables
(this is what initially hoookd me on links) and 2. the source code. I would love to do a side-by-side comparison of text-only browsers. Of course, this is not a topic anyone really cares about. I think links would be the fastest. Funny how he calls anyone who would use a text-only browser exclusively a "masochist". That is how I think of users who use the popular graphical browsers, waiting arbitray amounts of time on every website for pages to "load", making dozens of unnecessary connections to third party servers.
I haven't used Lynx in probably 15 years - this is my first time using it to submit a form, actually. I'm really enjoying it! I spend most of my day in Vim and Tmux otherwise. Thanks for sharing
Links also has a graphical interface, that displays images. It's still plenty fast. I love how instant web feels through links. It's like 1G fiber connection, 24GB RAM and 8 cores add up to something, but it's just as fast as it was almost 20 years ago, when I used it first time.
One of the projects I intent to one day do is a modern HTML-only browser with tabs, video support and other useful amenities.
I can play full games like krunker.io right now in browser with minimal lag.
I hope all the OS developers are taking note. OS might be just a browser plugin in the future instead of the other way round.
Humans have a brain that soaks new information like a sponge. Internet is the new crack. Having a limited browser is sad.
Edit: ok I tried it. I have to allow 2-3 cookies to run google.com. Facebook doesnot run. I think its a niche thing. Like among peers it would be fun to learn to run and run it but not a stable thing.
I think you don't understand the concept of an OS and a browser. If what you say happens, then your idea of a “browser” is actually the OS and the “plugin” is an application.
I just opened up krunker.io in Firefox. lol.
They seem to care a lot about my privacy, because the popup said "We care about your privacy". Then I clicked "Show purposes" and it listed ~20 cookies, six of which are "Always active". Ones like "Technically deliver ads or content" which seems super important. lol.
I can confirm the popup lag was minimal, I hope krunker.io developers note that. lol.
# If you really want the terminal's default colors, and if lynx is built using
# ncurses' default-color support, remove these two lines:
#normal: normal: lightgray:black
#default: normal: white:black
The Lynx styles stuff (and the initial cookie jar support) is definitely my most successful bit of code - made in 1996, still cropping up almost 25 years later.
The article reads much like an 'Ode to Lynx' without providing real reasons on why one should use it.
Yes, you can do some sort of accessibility testing with it (and I'm sure it is an useful tool for a lot more corner cases) but I was expecting a lot more after reading the title
Oh, hello there again! It's always good to see people testing their site on non-mainstream user-agents.
Some other terminal user-agents to try:
rdrview, go-readability: these are CLI utilities that use Mozilla's Readability algorithm to send HTML of just the article to stdout. Pipe that into w3m/lynx/links to read articles in the terminal. Testing in these ensures that your website will work in Firefox's Reader Mode. I also like to diff my website's HTML against the HTML output of these tools, just to see how much extraneous HTML I have.
Mozilla's Readability is written in JS; these tools are written in C and Go, respectively. With the exception of a few whitespace characters, their output of both seems to be identical.
If you want to go the extra mile and ensure that your website works with the most basic HTML parsers, try sending it through an HTML-to-Markdown converter that doesn't include inline html in the Markdown output. I test with pandoc's strict markdown output and make sure the resulting markdown is readable:
pandoc -f html -t markdown_strict https://example.com
None of this is a replacement for real a11y testing; most a11y technology leverages mainstream browsers (Chromium, FF, Safari, et cetera).
"Firefox seems to eat all 8GBs or my RAM lately so it's getting more tempting"
How many tabs are you opening? I have right here, right now 20 tabs opened in Firefox and Task Manager says my RAM is at 3.7 GB usage. That includes everything, from OS to other background apps that I'm running.
Firefox seems to never return any memory if you close tabs. My workaround for Firefox is using cgroups and limit memory for Firefox there, then it will not swap trash on my 24GB RAM system. Also every couple of weeks I restart it to reclaim the memory, or faster if needed.
Oh this is an interesting workaround. Didn't know you can limit memory to processes. Ubuntu doesn't handle out of memory at all. The whole system just locks up and freezes (at least on 20.04)
Like you say, it doesn't release memory so I need to restart periodically. If I forget then I need to do a hard restart when the system locks up (sometimes loosing work :/)
It seems strange.. it used to handle hundred of open tabs just fine and now it's completely crippled. My guess is that in making the browser faster they cache more and more - and Firefox devs prolly run on fancy machines with tons of RAM.
I think Firefox devs restart it many times a day during development, so they are not as affected. I think with 24GB of RAM I am at least on the edge of tons of memory qualifier and it still seems not enough after few weeks or intensive usage. Nevertheless it's baffling for me not to return the memory, but it's the default with small (roughly less than a page) allocations, because of how brk/sbrk work, Firefox should strive to do better.
I don't know what kind of PC and/or Firefox you have. I just test it by opening 20 more tabs on different heavy image sites (amazon, 9gag, reddit, twitter, instagram), with some having infinite scrolling and scrolled quite long. The memory jumped from 3.7 GB to 4.7 GB then closed them all. Task manager showed me the memory dropped down back to the original 3.7 GB.
I thought it's obvious is Windows, since I dropped Task Manager as a name there. Or to be absolutely purist, my main OS is Windows under VMWare's hypervisor ESXi, but I have all of them (MacOS, Linux, Android, BSD) as guests on different HDD's/partitions.
A salute to Lynx, I rely on it every day to cope with `text/html` e-mail in Mutt by virtue of the following line in my `mailcap` that has worked for anything hitting my inbox for about five years now:
Trying out the alternatives after reading the other comments here, perhaps I should give `w3m` a shot some day as it renders white space rather pleasantly?
75 comments
[ 4.0 ms ] story [ 168 ms ] threadMy blog's homepage is dominated by that construction, so my preferred alternative is w3m.
Lynx's keybindings never make sense to me.
I often open a news page or something and it's completely blank at first because so many scripts and assets got blocked by uMatrix. Before I actually unblock any domains, I try toggling reader mode on - a lot of times it's enough to let me read the plain HTML text which I actually wanted.
Slow and perhaps also unusable --- such as the popular anti-pattern of requiring JS to render static text, a great example of that being Twitter.
But I used to use it more often. Some years ago I worked mostly on a terminal window and by using lynx, I could browse the internet while everybody thought I was doing some serious work on the server (folks over there respected people working on black text-based screens)
But the killer app for me with web browsers is link hints, like in several browser extensions (e.g. vrome, tridactyl) and in browsers like qutebrowser. It takes a little getting used to at first to use efficiently, but I hate not having it when I use Lynx or w3m.
Slightly apropos, and for what it's worth, I've noticed that 99% of my browsing seems to be or start from maybe 20 or so web sites. I've started creating shell aliases to open those sites and there is no reason not to use the best browser for the site for each one. It works well with a philosophy like seems to be the one behind the surf browser, where you open each thing in its own window and use your window manager to deal with it, instead of using tabs.
and those shell aliases sound very similar to surfraw: https://en.wikipedia.org/wiki/Surfraw
Quite interesting to read Assange wrote Surfraw in 2000.
[0] https://en.wikipedia.org/wiki/ELinks#cite_note-2 [1] https://en.wikipedia.org/wiki/Links_(web_browser)#cite_note-...
That's my main use of Lynx these days - for ad-hoc extracting a bunch of links from a page, there's not much easier.
I've used both and links is vastly preferable, IMO.
Although I tend to use the name "elinks" since it's much more google-able.
elinks started out as an experimental fork of links but these days they’ve diverged a little. For example, links supports a whole plethora of operating systems, including DOS! While elinks is POSIX only.
One of the projects I intent to one day do is a modern HTML-only browser with tabs, video support and other useful amenities.
'void main() {}'
I can play full games like krunker.io right now in browser with minimal lag.
I hope all the OS developers are taking note. OS might be just a browser plugin in the future instead of the other way round.
Humans have a brain that soaks new information like a sponge. Internet is the new crack. Having a limited browser is sad.
Edit: ok I tried it. I have to allow 2-3 cookies to run google.com. Facebook doesnot run. I think its a niche thing. Like among peers it would be fun to learn to run and run it but not a stable thing.
lol?
> I hope all the OS developers are taking note. OS might be just a browser plugin in the future instead of the other way round.
Maybe the promise of "thin clients" might be realized. Anything is possible.
> Humans have a brain that soaks new information like a sponge.
Where did "Can't teach old dogs new tricks" come from then? Maybe fpr young children, but older people have trouble picking up new technology.
> I think its a niche thing.
I guess you are right - like a sponge.
I've started using Lagrange probably more than I ought to lately.
No thanks.
Yes, you can do some sort of accessibility testing with it (and I'm sure it is an useful tool for a lot more corner cases) but I was expecting a lot more after reading the title
I have at various times found myself unable to use a GUI, and these browsers saved my butt finding information to restore my system.
This is why I test all functionality in all three of them and ensure that they are just as able to use the site (or better) as Chrome Canary.
Some other terminal user-agents to try:
rdrview, go-readability: these are CLI utilities that use Mozilla's Readability algorithm to send HTML of just the article to stdout. Pipe that into w3m/lynx/links to read articles in the terminal. Testing in these ensures that your website will work in Firefox's Reader Mode. I also like to diff my website's HTML against the HTML output of these tools, just to see how much extraneous HTML I have.
Mozilla's Readability is written in JS; these tools are written in C and Go, respectively. With the exception of a few whitespace characters, their output of both seems to be identical.
=> https://github.com/eafer/rdrview rdrview
=> https://github.com/go-shiori/go-readability go-readability
Edbrowse is a line-mode browser that supports JavaScript via Duktape. Originally written for blind users, it's great for sighted users as well.
=> https://edbrowse.org/ edbrowse
If you want to go the extra mile and ensure that your website works with the most basic HTML parsers, try sending it through an HTML-to-Markdown converter that doesn't include inline html in the Markdown output. I test with pandoc's strict markdown output and make sure the resulting markdown is readable:
None of this is a replacement for real a11y testing; most a11y technology leverages mainstream browsers (Chromium, FF, Safari, et cetera).I occasionally use eww, though it a bit of unerogonomic pain. Esp without url autocomplete based on browser history.
Though I'm sure I could yak shave that with some elisp if I was motivated enough
(Firefox seems to eat all 8GBs or my RAM lately so it's getting more tempting)
How many tabs are you opening? I have right here, right now 20 tabs opened in Firefox and Task Manager says my RAM is at 3.7 GB usage. That includes everything, from OS to other background apps that I'm running.
Like you say, it doesn't release memory so I need to restart periodically. If I forget then I need to do a hard restart when the system locks up (sometimes loosing work :/)
It seems strange.. it used to handle hundred of open tabs just fine and now it's completely crippled. My guess is that in making the browser faster they cache more and more - and Firefox devs prolly run on fancy machines with tons of RAM.
I can record a video if you don't believe it.